A method for constructing an alarm information graph for power grid equipment

Abstract

The present invention discloses a method for constructing an alarm information graph for power grid equipment. The method includes constructing an alarm information database based on alarm information from power grid equipment and power grid systems; constructing a basic information database based on basic information from power grid equipment and power grid systems; constructing graph concept suspected relationships using a concept relationship tagging tool; constructing an alarm information-specific word segmentation tool based on the basic information database, and using the alarm information-specific word segmentation tool to segment the alarm information in the alarm information database to obtain alarm segmentation information; performing string forward clustering on the alarm segmentation information to obtain alarm information clustering results; constructing the alarm information into a knowledge graph based on the alarm information clustering results and graph concept suspected relationships using an analysis and display tool, and tagging the knowledge graph with alarm information to obtain an alarm information graph. This embodiment effectively obtains an alarm information graph for power grid equipment and improves the accuracy and precision of the alarm information graph.

Description

Technical Field

[0001] The present invention relates to the field of power grid knowledge graphs, and in particular to a method for constructing an alarm information graph for power grid equipment. Background Art

[0002] Currently, the power grid operates with a large number of devices and systems, each generating numerous alarms during operation. A single event can cause multiple systems and devices to generate the same alarm, along with numerous accompanying alarms, within the same timeframe. Without adequate processing and management tools, these alarms create a massive workload for production and maintenance personnel, potentially leading to missed critical and useful alarms, posing a serious threat to national security and public well-being. Knowledge graphs can imbue machines with perception and understanding capabilities, potentially enabling reasoning and decision-making, and are widely used in research related to inference and prediction. Currently, knowledge graphs are being combined with logical rules to implement embedded representation learning. These theoretical studies have contributed to improvements in the computational accuracy, efficiency, and fusion of inference results of knowledge graph reasoning techniques, laying a theoretical foundation for the application of big data in decision-making and inference.

[0003] The existing technology uses massive amounts of existing text data to provide a basis for intelligent diagnosis of equipment failures and constructs a knowledge graph. However, it lacks processing for batch alarm information and is unable to extract useful key alarm information, resulting in low graph accuracy and an inability to provide effective guidance for operation and maintenance personnel. Summary of the Invention

[0004] The present invention provides a method for constructing an alarm information map of power grid equipment, which can effectively obtain the alarm information map of power grid equipment and improve the accuracy and precision of the alarm information map.

[0005] In order to solve the above technical problems, an embodiment of the present invention provides a method for constructing an alarm information map of power grid equipment, comprising:

[0006] Build an alarm information database based on the alarm information of power grid equipment and power grid system; and build a basic information database based on the basic information of power grid equipment and power grid system;

[0007] Use the concept relationship marking tool to construct the suspected relationship of the graph concept;

[0008] Based on the basic information database, a proprietary word segmentation tool for alarm information is constructed. The alarm information in the alarm information database is segmented using the proprietary word segmentation tool to obtain alarm segmentation information. The alarm segmentation information is then subjected to string forward clustering to obtain alarm information clustering results.

[0009] According to the clustering results of the alarm information and the suspected relationship between the graph concepts, the alarm information is constructed into a knowledge graph through the analysis and display tool, and the knowledge graph is marked with alarm information to obtain the alarm information graph.

[0010] In implementing the embodiment of the present invention, an alarm information database is constructed based on the alarm information of the power grid equipment and the power grid system; and a basic information database is constructed based on the basic information of the power grid equipment and the power grid system; a graph concept suspected relationship is constructed through a concept relationship labeling tool; an alarm information-specific word segmentation tool is constructed based on the basic information database, and the alarm information in the alarm information database is segmented through the alarm information-specific word segmentation tool to obtain alarm segmentation information; the alarm segmentation information is forward clustered to obtain an alarm information clustering result; based on the alarm information clustering result and the graph concept suspected relationship, the alarm information is constructed as a knowledge graph through an analysis and display tool, and the knowledge graph is labeled with alarm information to obtain an alarm information graph. The alarm information clustering results are obtained through alarm information word segmentation and string forward clustering. The alarm information is constructed into a knowledge graph based on the suspected relationship of graph concepts and the alarm information clustering results. Batch alarm information is effectively processed, and useful key alarm information is classified and extracted. The alarm information graph of power grid equipment is effectively obtained, which improves the accuracy and precision of the alarm information graph, provides effective guidance for operation and maintenance personnel, and improves the safety of equipment within the power grid operation field.

[0011] As a preferred solution, an alarm information database is constructed based on the alarm information of power grid equipment and power grid system, specifically:

[0012] Obtain alarm information of power grid equipment and power grid system, and obtain all alarm information;

[0013] Replace the time and quantity in all alarm information with the time expression and quantity expression respectively to obtain the sorted alarm information;

[0014] The time expression is as follows:

[0015] ${yyyy-mm-dd hh:mm:ss}←year-month-day hour:minute:second;

[0016] ${yyyy-mm-dd hh:mm}←year-month-day hour:minute;

[0017] ${yyyy-mm-dd hh}←year-month-day-hour;

[0018] ${yyyy-mm-dd}←year-month-day;

[0019] Among them, the quantitative expression is specifically:

[0020] ${num}←quantity.

[0021] The sorted alarm information is constructed into an alarm information database.

[0022] As a preferred solution, a basic information database is constructed based on the basic information of power grid equipment and power grid system, specifically:

[0023] Obtain basic information corresponding to power grid equipment and power grid systems based on equipment ledger information and system ledger information database;

[0024] Build the basic information corresponding to the power grid equipment and power grid system into a basic information database;

[0025] The basic information corresponding to the power grid system includes the system name, the network topology structure where the system is located, and the system unique identifier;

[0026] The basic information corresponding to the power grid equipment includes the equipment name, the topology of the power grid system where the equipment is located, and the unique identification of the equipment.

[0027] As a preferred solution, the concept relationship marking tool is used to construct the suspected relationship of the graph concept, specifically:

[0028] Build a concept relationship tagging tool, and use it to edit concepts, concept corresponding attributes and corresponding relationships between concepts to obtain suspected relationships between concepts in the graph;

[0029] Among them, the concepts include systems, plants, equipment and alarm signals.

[0030] As a preferred solution, a dedicated word segmentation tool for alarm information is constructed based on the basic information database, specifically:

[0031] Extract the corresponding device name from the basic information database, and determine the concept type of the device based on the device name;

[0032] According to the concept type of the device, based on the Jieba word segmentation tool, add the corresponding device name and the unique identifier corresponding to the concept to build a proprietary word segmentation tool for alarm information.

[0033] As a preferred solution, the alarm segmentation information is subjected to string forward clustering to obtain the alarm information clustering results, specifically:

[0034] Sort the alarm segmentation information and select the first piece of information as the current alarm center point;

[0035] Circularly select the next piece of information and perform forward matching with the information of the current alarm center point to cyclically obtain the current alarm information clustering result; wherein, the forward matching is based on word segmentation;

[0036] Count the current alarm information clustering results corresponding to each piece of information to obtain the alarm information clustering results;

[0037] Among them, forward matching is specifically:

[0038] When the current alarm center point does not have different location information marked, the current information is matched with the current alarm center point. If they match, the current information and the current alarm center point are of the same type, and the different location information of the current alarm center point is recorded. The match means that the number of different terms in the two pieces of information does not exceed a preset threshold.

[0039] When the current alarm center point has different location information marked, the current information is compared with the information outside the other locations for term matching. If both match, the current information and the current alarm center point are of the same type, and the different location information of the current alarm center point is recorded;

[0040] During the term matching process, when the number of different terms in two pieces of information exceeds a preset threshold, the current information is used as another new alarm center point, and the new alarm center point is used as the current alarm center point.

[0041] As a preferred solution, based on the alarm information clustering results and the suspected relationship between graph concepts, the alarm information is constructed into a knowledge graph through analysis and display tools. Specifically:

[0042] Display the alarm information clustering results in thumbnails;

[0043] According to the suspected relationship of the graph concept, the alarm information of the alarm information clustering result is graphed through the analysis and display tool to obtain the current graph, and the inapplicable node relationships in the current graph are adjusted through the interface to generate a knowledge graph; in the process of generating the knowledge graph, all nodes are conflict detected.

[0044] As a preferred solution, the alarm information clustering results are displayed in a thumbnail format, specifically:

[0045] Display alarm information of different alarm center points and mark all the same alarm categories under different alarm center points;

[0046] Check all alarm information under all categories, find alarm information that does not belong to the current category, and mark the alarm information that does not belong to the current category as deleted;

[0047] Focus on the alarm center point with only a single message, re-segment the single message through manual word segmentation, and reclassify the single message.

[0048] As a preferred solution, after marking the map with alarm information to obtain the alarm information map, the following steps are further included:

[0049] When there is a change in the power grid information, the alarm information map is updated in real time according to the information change; the information change includes the exit of power grid equipment or power grid system, and the existence of new alarm information.

[0050] As a preferred solution, when there is a change in the power grid information, the alarm information map is updated in real time according to the information change, specifically:

[0051] If there is a power grid device or power grid system exit, the corresponding exit information is found on the alarm information map, and the exit information is marked to obtain a to-be-deleted information mark. According to the to-be-deleted information mark, it is uniformly deleted on the alarm information map; wherein, the exit information includes the exited power grid device, the exited power grid system and the exited alarm information;

[0052] If new alarm information appears, the new alarm information will be matched with the alarm information of each alarm center point respectively. When the new alarm information belongs to the same category as any alarm center point, the nodes, attributes and relationships corresponding to the new alarm information are generated according to the graph generation rules corresponding to the alarm information of the alarm center point of the same category as the new alarm information. According to the nodes, attributes and relationships corresponding to the new alarm information, the new alarm information is displayed in the alarm information graph; when the new alarm information does not belong to any category, a new alarm center point is constructed according to the new alarm information, and the corresponding nodes, attributes and relationships of the new cluster center point are constructed through the display interface. According to the corresponding nodes, attributes and relationships of the new cluster center point, the new alarm information is displayed in the alarm information graph. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 : A flow chart of an embodiment of a method for constructing an alarm information map of power grid equipment provided by the present invention;

[0054] Figure 2 : A simplified schematic diagram of constructing a map of an embodiment of a method for constructing an alarm information map of a power grid device provided by the present invention. DETAILED DESCRIPTION

[0055] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0056] Example 1

[0057] Please refer to Figure 1, which is a flow chart of a method for constructing an alarm information graph for power grid equipment provided by an embodiment of the present invention. This embodiment obtains alarm information clustering results through alarm information word segmentation and string forward clustering. Based on the suspected relationship between graph concepts and the alarm information clustering results, the alarm information is constructed into a knowledge graph, thereby effectively obtaining an alarm information graph for power grid equipment and improving the accuracy and precision of the alarm information graph. The alarm information graph construction method includes steps 101 to 105, each of which is specifically as follows:

[0058] Step 101: construct an alarm information database based on the alarm information of the power grid equipment and the power grid system; and construct a basic information database based on the basic information of the power grid equipment and the power grid system.

[0059] In this embodiment, a simplified schematic diagram of the atlas is constructed, such as Figure 2 As shown, an alarm information database based on device alarms and system alarms is constructed, and the alarm information database includes the device alarm information database and the system alarm information database; a basic information database for devices and systems is constructed, and the basic information database includes the device basic information database and the system basic information database; concepts in the knowledge graph and possible relationships between concepts (suspected relationships between graph concepts) are constructed; a word segmentation tool based on the device and system basic information database is constructed; the word segmentation tool is used to segment the alarm information; a string forward clustering algorithm is constructed to cluster all segmented alarm information; a clustering result and analysis display tool is constructed to construct a graph for the confirmed alarm information with one click, and the alarm information is marked; when new alarm information appears or a device or system exits, the word segmentation and clustering are re-performed, and the graph is automatically updated according to the marking.

[0060] Optionally, step 101 specifically includes steps 1011 to 1012, and each step is specifically as follows:

[0061] Step 1011: Acquire alarm information of power grid equipment and power grid system, and obtain all alarm information;

[0062] Replace the time and quantity in all alarm information with the time expression and quantity expression respectively to obtain the sorted alarm information;

[0063] The time expression is as follows:

[0064] ${yyyy-mm-dd hh:mm:ss}←year-month-day hour:minute:second;

[0065] ${yyyy-mm-dd hh:mm}←year-month-day hour:minute;

[0066] ${yyyy-mm-dd hh}←year-month-day-hour;

[0067] ${yyyy-mm-dd}←year-month-day;

[0068] Among them, the quantitative expression is specifically:

[0069] ${num}←quantity;

[0070] The sorted alarm information is constructed into an alarm information database.

[0071] In this embodiment, all alarm information is collected and sorted, and the time, quantity and other words in the alarm information are replaced with corresponding time or quantity expressions to complete the construction of an alarm information database based on device alarms and system alarms.

[0072] Step 1012: Based on the equipment inventory information and the system inventory information library, obtain the basic information corresponding to the power grid equipment and the power grid system; construct the basic information corresponding to the power grid equipment and the power grid system into a basic information library; wherein, the basic information corresponding to the power grid system includes the system name, the network topology structure where the system is located, and the system unique identifier; wherein, the basic information corresponding to the power grid equipment includes the device name, the power grid system topology structure where the device is located, and the device unique identifier.

[0073] In this embodiment, basic information about devices and systems is collected through the device and system inventory databases to build a basic information database. The database includes information about systems, including the system name, the network topology in which the system resides, and the system's unique identifier; and information about devices, including the device name, the network topology in which the device resides, and the device's unique identifier.

[0074] Step 102: Use the concept relationship tagging tool to construct the suspected relationship of the graph concept.

[0075] Optionally, step 102 is specifically as follows: constructing a concept relationship marking tool, and using the concept relationship marking tool to edit concepts, concept corresponding attributes and corresponding relationships between concepts to obtain suspected relationships of graph concepts; wherein, concepts include systems, plants, equipment and alarm signals.

[0076] In this embodiment, a concept relationship tagging tool is constructed, which can realize the editing of concepts, editing of corresponding attributes of concepts, and editing of corresponding relationships between concepts. Specific concepts include: system, plant, equipment, alarm signal and other concepts.

[0077] It should be noted that the concept relationship tagging tool uses Java development tools and is mainly used to manage concepts and concept relationships through an interface.

[0078] Step 103: Based on the basic information database, a dedicated word segmentation tool for alarm information is constructed, and the alarm information in the alarm information database is segmented using the dedicated word segmentation tool to obtain alarm segmentation information, and the alarm segmentation information is forward clustered to obtain alarm information clustering results.

[0079] In this embodiment, a word segmentation tool based on the equipment and system basic information library is constructed, and the alarm information is segmented. A string forward clustering algorithm is constructed to cluster all segmented alarm information (alarm segmentation information) to obtain the alarm information clustering result.

[0080] Optionally, step 103 specifically includes steps 1031 to 1033, each of which is as follows:

[0081] Step 1031: Extract the corresponding device name from the basic information database, and determine the concept type to which the device belongs based on the device name; based on the concept type to which the device belongs, add the corresponding device name and the unique identifier corresponding to the concept based on the Jieba word segmentation tool to build a proprietary word segmentation tool for alarm information.

[0082] In this embodiment, the corresponding device names are extracted from the basic information library of the constructed devices and systems, the concept type to which the devices belong is defined, and based on the Jieba word segmentation tool, the unique identifiers corresponding to the corresponding device names and concepts are added to construct a proprietary word segmentation tool for alarm information.

[0083] Step 1032: Segment the alarm information in the alarm information database using a dedicated segmentation tool for alarm information to obtain alarm segmentation information.

[0084] In this embodiment, all collected alarm information is segmented using a dedicated segmentation tool for alarm information.

[0085] Step 1033: sort the alarm segmentation information and select the first information as the current alarm center point;

[0086] Circularly select the next piece of information and perform forward matching with the information of the current alarm center point to cyclically obtain the current alarm information clustering result; wherein, the forward matching is based on word segmentation;

[0087] Count the current alarm information clustering results corresponding to each piece of information to obtain the alarm information clustering results;

[0088] Among them, forward matching is specifically:

[0089] When the current alarm center point does not have different location information marked, the current information is matched with the current alarm center point. If they match, the current information and the current alarm center point are of the same type, and the different location information of the current alarm center point is recorded. The match means that the number of different terms in the two pieces of information does not exceed a preset threshold.

[0090] When the current alarm center point has different location information marked, the current information is compared with the information outside the other locations for term matching. If both match, the current information and the current alarm center point are of the same type, and the different location information of the current alarm center point is recorded;

[0091] During the term matching process, when the number of different terms in two pieces of information exceeds a preset threshold, the current information is used as another new alarm center point, and the new alarm center point is used as the current alarm center point.

[0092] In this embodiment, the clustering algorithm steps are as follows:

[0093] 1) Set a threshold, that is, a preset threshold. If the threshold is exceeded, it is considered not to be the same category;

[0094] 2) Sort the alarms and select the first alarm as the center (alarm center point);

[0095] 3) Circularly select the next alarm information and perform forward matching with the alarm information in the alarm center point in units of word segments;

[0096] If the alarm center point doesn't have different location information marked, the alarm information is directly matched. If the number of different terms in the two alarms does not exceed the threshold, the current records are considered to be of the same type and the different location information of the center point is recorded. It should be noted that the same word segmentation tool is used, and the marking position will not change.

[0097] When the alarm information is marked with different position information, the next alarm information needs to be compared with the terms other than those in the other positions. If they match, they are considered to be of the same type. It should be noted that a match means that the number of different terms in the two alarm information does not exceed a threshold.

[0098] During the matching process, if the number of different entries exceeds the threshold, it is considered to be a new type of alarm information center point, and the current alarm information is used as another alarm information center point.

[0099] 4) Temporarily store all clustering results.

[0100] Step 104: Based on the alarm information clustering results and the suspected relationship between the graph concepts, the alarm information is constructed into a knowledge graph through the analysis and display tool, and the knowledge graph is marked with alarm information to obtain an alarm information graph.

[0101] Optionally, based on the alarm information clustering results and the suspected relationships between graph concepts, the alarm information is constructed into a knowledge graph through an analysis and display tool, specifically including steps 41 and 42 as follows:

[0102] Step 41: Display the alarm information clustering results in thumbnails;

[0103] Optionally, step 41 is specifically as follows: display the alarm information of different alarm center points, and mark all the same alarm categories under different alarm center points; check the alarm information under all categories, find the alarm information that does not belong to the current category, and remove the alarm information that does not belong to the current category; focus on the alarm center point with only a single message, re-segment the single message through manual segmentation, and reclassify the single message.

[0104] In this embodiment, the clustering results are briefly displayed:

[0105] (1) Display the alarm information of all different center points and mark all the same alarm categories under the corresponding center points; (2) Check the alarm information under all categories to see if there are any that do not belong to the current category, and remove the alarms that do not belong to the current category; (3) Focus on the alarms with only a single message, and re-segment the alarms using manual segmentation to reclassify them.

[0106] Step 42: Based on the suspected relationship of the graph concept, the alarm information of the alarm information clustering result is graphed through the analysis and display tool to obtain the current graph, and the node relationships that are not applicable in the current graph are adjusted through the interface to generate a knowledge graph; wherein, in the process of generating the knowledge graph, all nodes are conflict detected.

[0107] In this embodiment, all clustering result warning information is graphed based on the constructed concepts and the relationships between concepts (graph concept suspected relationships). Inappropriate node relationships are adjusted through the interface, and a graph is generated with one click. During the graph generation process, all nodes are conflict checked to prevent the reconstruction of nodes and the duplication of existing nodes.

[0108] Step 105: When there is a change in the power grid information, the alarm information map is updated in real time according to the information change; the information change includes the exit of power grid equipment or power grid system, and the existence of new alarm information.

[0109] Optionally, step 105 is specifically as follows:

[0110] When a grid device or grid system exits the grid, the corresponding exit information is found on the alarm information map, and the exit information is marked to obtain a to-be-deleted information mark. Based on the to-be-deleted information mark, the exit information is uniformly deleted on the alarm information map; wherein, the exit information includes the exited grid device, the exited grid system, and the exited alarm information;

[0111] In this embodiment, when a new device or system exits, the corresponding device, system, and alarm information are found on the map, marked, and then deleted in a unified manner.

[0112] When new alarm information appears in the power grid, the new alarm information is matched with the alarm information of each alarm center point respectively. When the new alarm information belongs to the same category as any alarm center point, the nodes, attributes and relationships corresponding to the new alarm information are generated according to the graph generation rules corresponding to the alarm information of the alarm center point of the same category as the new alarm information. According to the nodes, attributes and relationships corresponding to the new alarm information, the new alarm information is displayed in the alarm information graph; when the new alarm information does not belong to any category, a new alarm center point is constructed according to the new alarm information, and the nodes, attributes and relationships corresponding to the new cluster center point are constructed through the display interface. According to the nodes, attributes and relationships corresponding to the new cluster center point, the new alarm information is displayed in the alarm information graph.

[0113] In this embodiment, when a new alarm occurs, it is matched with the alarm information of the previous alarm center. If they belong to the same category, the graph generation rules corresponding to the center alarm information are used to generate the corresponding nodes, attributes, and relationships. If they do not belong to any category, a new cluster center is constructed, and the corresponding nodes, attributes, and relationships are constructed using the display interface.

[0114] In implementing the embodiment of the present invention, an alarm information database is constructed based on the alarm information of the power grid equipment and the power grid system; and a basic information database is constructed based on the basic information of the power grid equipment and the power grid system; a graph concept suspected relationship is constructed through a concept relationship labeling tool; an alarm information-specific word segmentation tool is constructed based on the basic information database, and the alarm information in the alarm information database is segmented through the alarm information-specific word segmentation tool to obtain alarm segmentation information; the alarm segmentation information is forward clustered to obtain an alarm information clustering result; based on the alarm information clustering result and the graph concept suspected relationship, the alarm information is constructed as a knowledge graph through an analysis and display tool, and the knowledge graph is labeled with alarm information to obtain an alarm information graph. The alarm information clustering results are obtained through alarm information word segmentation and string forward clustering. The alarm information is constructed into a knowledge graph based on the suspected relationship of graph concepts and the alarm information clustering results. Batch alarm information is effectively processed, and useful key alarm information is classified and extracted. The alarm information graph of power grid equipment is effectively obtained, which improves the accuracy and precision of the alarm information graph, provides effective guidance for operation and maintenance personnel, and improves the safety of equipment within the power grid operation field.

[0115] The above specific embodiments further illustrate the objectives, technical solutions, and beneficial effects of the present invention. It should be understood that the above are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention for those skilled in the art.

Claims

1. A method for constructing an alarm information map of power grid equipment, characterized in that: include: Build an alarm information database based on the alarm information of power grid equipment and power grid system; and constructing a basic information database based on the basic information of the power grid equipment and the power grid system; Use the concept relationship marking tool to construct the suspected relationship of the graph concept; Based on the basic information database, a dedicated word segmentation tool for alarm information is constructed, and the alarm information in the alarm information database is segmented by the dedicated word segmentation tool to obtain alarm segmentation information, and the alarm segmentation information is subjected to string forward clustering to obtain an alarm information clustering result; Based on the alarm information clustering results and the suspected relationship between the graph concepts, the alarm information is constructed into a knowledge graph through an analysis and display tool, and the knowledge graph is marked with alarm information to obtain an alarm information graph; The alarm segmentation information is subjected to character string forward clustering to obtain the alarm information clustering result, which is specifically: Sort the alarm segmentation information and select the first information as the current alarm center point; Circularly select the next piece of information and perform forward matching with the information of the current alarm center point, and cyclically obtain the current alarm information clustering result; wherein, the forward matching is based on word segmentation; Count the current alarm information clustering results corresponding to each piece of information to obtain the alarm information clustering results; The forward matching is specifically as follows: When the current alarm center point does not have different location information marked, the current information is matched with the current alarm center point. If they match, the current information and the current alarm center point are of the same type, and the different location information of the current alarm center point is recorded; wherein the match means that the number of different terms in the two pieces of information does not exceed a preset threshold; When the current alarm center point has different location information marked, the current information is compared with the information outside the other locations for term matching. If both match, the current information and the current alarm center point are of the same type, and the different location information of the current alarm center point is recorded; During the term matching process, when the number of different terms in two pieces of information exceeds the preset threshold, the current information is used as another new alarm center point, and the new alarm center point is used as the current alarm center point.

2. The method for constructing an alarm information graph of a power grid device according to claim 1, characterized in that: The alarm information database is constructed based on the alarm information of the power grid equipment and the power grid system, specifically: Acquire alarm information of the power grid equipment and the power grid system, and obtain all alarm information; Replacing the time and quantity in all the alarm information with the time expression and the quantity expression respectively to obtain the sorted alarm information; The time expression is specifically: ${yyyy-mm-dd hh:mm:ss}←year-month-day hour:minute:second; ${yyyy-mm-dd hh:mm}←year-month-day hour:minute; ${yyyy-mm-dd hh}←year-month-day-hour; ${yyyy-mm-dd}←year-month-day; The quantitative expression is specifically: ${num}←quantity; The organized alarm information is constructed into the alarm information database.

3. The method for constructing an alarm information graph of a power grid device according to claim 1, wherein: The basic information database is constructed based on the basic information of the power grid equipment and the power grid system, specifically: Obtain basic information corresponding to the power grid equipment and the power grid system based on the equipment ledger information and the system ledger information library; Constructing the basic information corresponding to the power grid equipment and the power grid system into the basic information database; The basic information corresponding to the power grid system includes the system name, the network topology structure where the system is located, and the system unique identifier; The basic information corresponding to the power grid equipment includes the equipment name, the topology of the power grid system where the equipment is located, and the unique identification of the equipment.

4. The method for constructing an alarm information graph of a power grid device according to claim 1, wherein: The concept relationship marking tool is used to construct the suspected relationship of the graph concept, specifically: Constructing the concept relationship marking tool, and editing the concepts, corresponding attributes of concepts and corresponding relationships between concepts through the concept relationship marking tool to obtain the suspected relationship of the graph concepts; The concepts include systems, plants, equipment and alarm signals.

5. The method for constructing an alarm information graph of a power grid device according to claim 1, wherein: The method of constructing a dedicated word segmentation tool for alarm information based on the basic information database is as follows: Extracting a corresponding device name from the basic information database, and determining the concept type to which the device belongs based on the device name; According to the concept type to which the device belongs, based on the Jieba word segmentation tool, the corresponding device name and the unique identifier corresponding to the concept are added to construct the alarm information proprietary word segmentation tool.

6. The method for constructing an alarm information graph of a power grid device according to claim 1, wherein: According to the clustering results of the alarm information and the suspected relationship between the graph concepts, the alarm information is constructed into a knowledge graph through an analysis and display tool, specifically: Displaying the alarm information clustering results in thumbnails; According to the suspected relationship of the graph concepts, the alarm information of the alarm information clustering result is graphed through the analysis and display tool to obtain the current graph, and the inapplicable node relationships in the current graph are adjusted through the interface to generate the knowledge graph; wherein, in the process of generating the knowledge graph, all nodes are subjected to conflict detection.

7. The method for constructing an alarm information graph of a power grid device according to claim 6, characterized in that: The clustering result of the alarm information is briefly displayed, specifically: Display alarm information of different alarm center points and mark all the same alarm categories under the different alarm center points; Check the alarm information under all categories, find the alarm information that does not belong to the current category, and remove the alarm information that does not belong to the current category; Focus on checking the alarm center point with only a single piece of information, re-segment the single piece of information by manual word segmentation, and reclassify the single piece of information.

8. The method for constructing an alarm information graph of a power grid device according to claim 1, wherein: After marking the map with alarm information to obtain an alarm information map, the method further includes: When there is information change in the power grid, the alarm information map is updated in real time according to the information change; wherein the information change includes the exit of power grid equipment or power grid system, and the existence of new alarm information.

9. The method for constructing an alarm information graph of a power grid device according to claim 8, characterized in that: When the power grid information changes, the alarm information map is updated in real time according to the information changes, specifically: When the grid device or grid system exits, the corresponding exit information is found on the alarm information map, and the exit information is marked to obtain a to-be-deleted information mark, and the exit information is uniformly deleted on the alarm information map according to the to-be-deleted information mark; wherein the exit information includes the exited grid device, the exited grid system, and the exited alarm information; When the power grid has the situation where new alarm information exists, the new alarm information is matched with the alarm information of each alarm center point respectively. When the new alarm information belongs to the same category as any of the alarm center points, the nodes, attributes and relationships corresponding to the new alarm information are generated according to the graph generation rules corresponding to the alarm information of the alarm center point of the same category as the new alarm information. According to the nodes, attributes and relationships corresponding to the new alarm information, the new alarm information is displayed in the alarm information graph; when the new alarm information does not belong to any category, a new alarm center point is constructed according to the new alarm information, and the nodes, attributes and relationships corresponding to the new cluster center point are constructed through the display interface. According to the nodes, attributes and relationships corresponding to the new cluster center point, the new alarm information is displayed in the alarm information graph.

Images